Figure 2
Results from the opt-in/out block. (a) Probability of opting out as a function of the task difficulty and reward for the opt-out choice. Participants opted out more frequently when the task was easier, and the reward was larger. The error bars indicate the standard errors of the means across participants. (b) Effects of the cognitive parameters on participants’ decision to opt out. The predictors affected the opt-out choice positively, but only for confidence did the 95% credible interval (CI) not contain zero. The points indicate the medians of the posterior distributions, and the error bars indicate the 50, 80, and 95% CIs. (c) Decision accuracy as a function of the task difficulty and reward for the opt-out choice: Majority decisions by voluntary participation versus loners’ individual decisions. Although the majority decisions by voluntary participation were worse than the loners’ individual decisions when the task was easier, the majority decisions began to outperform the individual decisions as the task got harder. The resampled data set in which no one voted was excluded from calculating the accuracy. (d) Accuracy of majority decisions as a function of the task difficulty and reward for the opt-out choice: Voluntary participation versus mandatory participation versus unbiased majority. The unbiased majority decisions were more accurate than the voluntary participation when the task was easy, but the difference in accuracy almost diminished as the task became difficult. For the mandatory participation, the accuracy is collapsed across the rewards of the opt-out choice because the 25 participants were treated as if all had chosen to opt in regardless of the reward.
